Integration

GoodMem integration

GoodMem is a memory layer for AI agents that you run yourself or as a GoodMem Cloud instance. Create spaces, store text or documents in them as memories (chunked and embedded in the background), list and fetch memories and their page images, and run semantic retrieval across spaces, optionally reranked and summarized by a configured LLM. Also read the embedder, LLM, reranker and decision-model configurations, ask a decision model structured questions, and run layout-aware OCR on a document. Not exposed: creating or changing embedders, LLMs, rerankers and decision models (their requests carry upstream provider credentials, and their credentials are never returned here), identity and access administration (users and enrollments, service identities, API keys, access-policy grants and role assignments), server administration (drain, license reload, job purge, retrieval log policies, ownership transfers, system initialization), endpoint pings and the HEAD variant of the content download. Connect with your server's REST base URL and an API key created in the GoodMem console or with the goodmem CLI, sent in the x-api-key header.

Connect GoodMem How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in GoodMem.Built from the 29 GoodMem tools below, on the connection you add once.

What you can do with GoodMem

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Memory, 10 operations

batch_create_memoriesBatch create memoriesPOSTCreates several memories in one call; each item reports its own success or error. Items take the same fields as create_memory.
batch_delete_memoriesBatch delete memoriesPOSTPermanently deletes memories selected by id or by a filter inside a space; each selector reports its result. Cannot be undone.
batch_get_memoriesBatch get memoriesPOSTReturns several memories by id in one call; each item reports its own success or error.
create_memoryCreate memoryPOSTStores text or base64 content as a memory in a space; it starts PENDING while chunking and embedding run in the background (check processingStatus with get_memory).
delete_memoryDelete memoryDELETEPermanently deletes a memory and its chunks. Cannot be undone.
download_memory_contentDownload memory contentGETDownloads a memory's original content as a file. Answers 404 when the content is stored externally (see originalContentRef in get_memory).
get_memoryGet memoryGETReturns one memory: space, content type, processing status (PENDING, PROCESSING, COMPLETED, FAILED), metadata and page-image status; includeContent adds the original content.
list_space_memoriesList space memoriesGETLists the memories of one space with their processing status and metadata, optionally filtered by status or a metadata filter expression. Page with nextToken.
retrieve_memoriesRetrieve memoriesGETSearches the memories of one or more spaces for a query and returns the matching chunks with relevance scores, plus an LLM answer when postProcessorLlmId is set.
retrieve_memories_advancedRetrieve memories advancedPOSTSearches spaces with per-space filters and embedder weights, context items, HNSW tuning and a post-processor. Returns chunks with relevance scores, plus an LLM answer when one runs.

Space, 5 operations

create_spaceCreate spacePOSTCreates a space: a container of memories embedded with one or more embedders (ids from list_embedders) and a default chunking strategy. Returns the space with its spaceId.
delete_spaceDelete spaceDELETEPermanently deletes a space with all its memories, chunks and embedder links. Cannot be undone.
get_spaceGet spaceGETReturns one space: name, labels, embedders, default chunking, owner and timestamps (milliseconds since epoch).
list_spacesList spacesGETLists the spaces your key can read, with their embedders, labels and chunking settings; filter by owner, name glob or labels. Page with nextToken.
update_spaceUpdate spacePUTRenames a space or changes its labels; fields left out stay unchanged. Embedders, chunking and owner cannot be changed. Returns the updated space.

Decision model, 3 operations

ask_decision_modelAsk decision modelPOSTSends named questions about some content to a decision model and returns its answers. Each call is one provider request that may incur charges: do not retry it automatically.
get_decision_modelGet decision modelGETReturns one decision model's settings and capabilities (question types, media, limits). Credentials are not returned.
list_decision_modelsList decision modelsGETLists the decision models (Jev, Laya, Clef providers) your key can read, newest first, with their supported question types. Page with nextPageToken. Credentials are not returned.

Embedder, 2 operations

get_embedderGet embedderGETReturns one embedder configuration: provider, model, dimensionality, endpoint, modalities and labels. Credentials are not returned.
list_embeddersList embeddersGETLists the embedder configurations your key can read: provider, model, dimensionality, endpoint and labels. Use an embedderId in create_space. Credentials are not returned.

Llm, 2 operations

get_llmGet llmGETReturns one LLM configuration: provider, model, capabilities and default sampling. Credentials are not returned.
list_llmsList llmsGETLists the LLM configurations your key can read: provider, model, capabilities, default sampling and labels. Use an llmId as a retrieval post-processor. Credentials are not returned.

Memory page, 2 operations

download_memory_page_imageDownload memory page imageGETDownloads one page image of a document memory as a file. Give dpi or contentType when the page has several renditions.
list_memory_pagesList memory pagesGETLists the page images extracted from a document memory (page index, DPI, MIME type, size), optionally limited to a page range. Page with nextToken.

Reranker, 2 operations

get_rerankerGet rerankerGETReturns one reranker configuration: provider, model, endpoint and labels. Credentials are not returned.
list_rerankersList rerankersGETLists every reranker configuration your key can read (no pagination), newest first. Use a rerankerId as a retrieval post-processor. Credentials are not returned.

Document, 1 operation

ocr_documentOcr documentPOSTRuns layout-aware OCR on a base64 document or image (PDF, TIFF, PNG, JPEG, BMP) and returns per-page text cells with bounding boxes, optionally Markdown.

Instance, 1 operation

get_instanceGet instanceGETReturns the GoodMem instance's id, owner and audit timestamps. Requires READ_INSTANCE (the ADMIN role).

System, 1 operation

get_system_infoGet system infoGETReturns the server's version, git commit, build time and capability flags. Needs no key.

Use GoodMem from Claude Code, Cursor or Codex

Connect GoodMem in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its GoodMem operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect GoodMem

Create your workspace
Sign up free. Nothing to install.
Connect GoodMem once
Add your GoodMem connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.